Zackriya-Solutions/meetily · info · anyhow::Error

Retranscription cancelled

Error message

Retranscription cancelled

What it means

Cooperative cancellation checkpoint: before the CPU-heavy decode stage, run_retranscription checks the RETRANSCRIPTION_CANCELLED AtomicBool and bails out if the user cancelled. The flag is set only by cancel_retranscription and reset at job start, so this error always reflects an intentional cancel, not a malfunction.

Source

Thrown at frontend/src-tauri/src/audio/retranscription.rs:196

    provider: Option<String>,
) -> Result<RetranscriptionResult> {
    let folder_path = PathBuf::from(&meeting_folder_path);
    let audio_path = find_audio_file(&folder_path)?;

    // Determine which provider to use (default to whisper)
    let use_parakeet = provider.as_deref() == Some("parakeet");

    info!(
        "Starting retranscription for meeting {} with language {:?}, model {:?}, provider {:?}",
        meeting_id, language, model, provider
    );

    // Emit progress: decoding
    emit_progress(&app, &meeting_id, "decoding", 5, "Decoding audio file...");

    // Check for cancellation
    if RETRANSCRIPTION_CANCELLED.load(Ordering::SeqCst) {
        return Err(anyhow!("Retranscription cancelled"));
    }

    // Decode the audio file (CPU-intensive, run in blocking task)
    let path_for_decode = audio_path.clone();
    let decoded = tokio::task::spawn_blocking(move || {
        decode_audio_file(&path_for_decode)
    })
    .await
    .map_err(|e| anyhow!("Decode task panicked: {}", e))??;
    let duration_seconds = decoded.duration_seconds;

    info!(
        "Decoded audio: {:.2}s, {}Hz, {} channels",
        duration_seconds, decoded.sample_rate, decoded.channels
    );

    emit_progress(&app, &meeting_id, "decoding", 15, "Converting audio format...");

View on GitHub (pinned to 0281737d87)

Solutions

1. Treat this as expected control flow: show 'Cancelled' in the UI, not an error
2. Restart the retranscription if the cancellation was accidental
3. Emit a terminal cancelled progress event so the UI resets its state cleanly

Example fix

// before
let result = run_retranscription(app, meeting_id, folder, lang, model, provider).await?;

// after - branch on cancellation so it is not surfaced as a failure
match run_retranscription(app, meeting_id, folder, lang, model, provider).await {
    Ok(result) => Ok(result),
    Err(e) if e.to_string().contains("cancelled") => {
        emit_progress(&app, &meeting_id, "cancelled", 0, "Cancelled by user");
        Err(e)
    }
    Err(e) => Err(e),
}

In the retranscription result handler, branch on the error message: 'Retranscription cancelled' maps to a neutral 'Cancelled' UI state (no error toast, reset progress), all other errors map to failure with retry option.

When it happens

Trigger: User clicks Cancel (cancel_retranscription_command) while a retranscription job is between the 'decoding' progress emission and the spawn_blocking decode call.

Common situations: Cancelling a long retranscription of a long meeting because the wrong meeting, language, or model was chosen.

Related errors

• VAD processing cancelled
• Import cancelled
• Retranscription already in progress
• No audio file found in: {}
• No speech detected in audio file

AI-assisted analysis of Zackriya-Solutions/meetily@0281737d87 (2026-08-16). Data as JSON: /api/errors/b2c2c13eed53f71f. Report an issue: GitHub.